Advances in solution methods for optimisation of multiple quality characteristics in manufacturing processes

Abstract

A typical problem generally encountered in the quality control and improvement operations of manufacturing processes involves simultaneously optimising multiple critical quality characteristics (or 'multiple responses'). These type of problems are so-called 'multiple response optimisation (MRO) problems'. Owing to correlation between multiple responses, trade-off solution(s) are inevitable. The term 'trade-off' is an explicit compromised solution, considering the bias from the targets and variability in the responses. The global best solution for such a problem is usually unknown. Over the years, various solution methods and their theoretical advancement are proposed. However, only a handful of critical reviews are evident in open literature. Available review articles on MRO seem insufficient and address only specific aspects of the solution methods (e.g., response-surface modelling, problem formulation, or optimisation techniques). In this study, five different types of solution methods suggested for MRO problems are categorised and critically reviewed, including response-surface-based contour plot, response surface-free data mining methods, etc. The theoretical relationship between the MRO and the multi-objective-optimisation solution methods is analysed along with identifying potential research direction in this domain.